Adapting a Computer PSU

[Crackle]

Looking at the circuit around the TL594 I dont think I will be able to do any thing until I get some specific help and guidance as the circuits are so small. With my poor eyesight it would not be suitable for me to experiment. So Ill wait to see if Mark can come up with the most sensible places to make the required modifications.
I have seen warnings that some capacitors may need changing to higher voltage ones if the output voltage is taken too high. It looks like the lowest voltage cap is 16v. 15 volts would be more than enough for my needs on this project.
What is that largish black rectangular surface mounted thing near the bottom left corner of the TL594, it has a white line across one end. Is it an electrolytic?

Mike

[Mark Hennessy]

The PS25A is a combined PFC and main switcher IC. In other words, it can do the whole lot. So I'm confused by the presence of the TL594. It's possible they decided to only use half of the PS25A - the PFC half - but if so, why not just use a chip that does PFC only? There's no shortage of them.

The standby supply has its own controller on the main PCB (the DIL 7 package), so it's not related to that.

If this was a supply that made other rails using separate DC-DC converters for each (as I described on that other thread), then the TL595 would have something to do - though there are better choices for secondary-side buck converters. Noting that the large toroidal inductor on the main PCB has at least 2 windings, it's possible it's being used as a mag-amp to further improve the regulation, but that would be unusual for a single rail supply.

I really need to get my hands on one of these to find out what's going on - I find it hard to process it all from photos - even good ones like these. I have a DC7800 SFF here, but that uses a different PSU. I'll take it apart later, as the innards could be the same or very similar, but IIRC that one is a conventional multi-output one. Failing that, I'll borrow one from the old machines at work that I mentioned earlier...

[Nick]

The PS25A is a combined PFC and main switcher IC. In other words, it can do the whole lot. So I'm confused by the presence of the TL594...

I think that's the CM6802 - the PS25A is a power quality monitor.

But, yes, why have two PWM controllers...

[Mark Hennessy]

Yeah, brain fart.

Looking more closely - now that I'm more awake after a bike ride with my son (who's nearly as fast as me, and it won't be long before he is) - I reckon the primary converter is definitely being controlled by the TL594. The transformer between the primary cap and the optocouplers is carrying the base or gate drive from the secondary side to the primary side. I would have spotted that sooner if I'd had the object in my hand - I'm not very good at seeing in 3D from the computer screen. But why are there 2 opto-couplers feeding back from secondary to primary? Obviously one must be for the standby supply, but if the TL594 is on the secondary side, why is there feedback back to the primary? I don't think it's to tell the PFC IC to throttle back in standby because it should do that natively (which is no-doubt why the SB supply is a bit lower when the main converter is not running - not that that matters, as the motherboard will be converting it down locally).

It's tempting to put it down as a strange or lazy choice, but in my experience it's best not to be such a smart-arse - there are often new tricks to be learnt from seemingly unconventional setups... But it is entirely possible that half of that primary-side chip is just not being utilised...

Something else on at the mo, but will see if I can make sense of the TL594 PCB later. Will be a lot easier with one in front of me, so might have to wait until next week.

Mike, a thought occurs before we get too far down this road - will it be quiet enough for the intended purpose? Even though it's only giving 12V, it should be easy enough to check at this point, before too much more effort is spent on further investigations...

[Crackle]

Hi Mark, yes I can test that.

Mike

[Nick]

I suspect one optocoupler is to take the PS25A power good signal back to the primary and the second one is from the error amp for the +5VSB. The TL594 should be on the primary side. I suspect this is not that an unusual design  although HP do odd stuff with connectors etc.

[Crackle]

I tested the PSU and there was some noise on receive, but this mostly went when I reconnected the PSU earth link, which I had lifted. Transmission reports come back with clear mod and no background noise, may need to confirm this with more local reports.
Also no indication of breakthrough on FM which I was getting before with the commercial Ham type PSU I bought and sent back.  The breakthrough on that was cured when I isolated the earth to the PSU, but I didn't wish to run the PSU with no earth and didn't want to invalidate warrantee by adapting the internal circuit to remove the mains earth from the DC output negative.

But I am happy to say the PSU seems suitable from RFI point of view, I think further improvement can be made on receive by earthing the PSU to the transceiver, better screening, and or adding more common mode filtering on the OP.
Its interesting to note that when the PSU is turned on there is a brief increase in noise on receive which goes after about 3 seconds.
Thanks
Mike

[Mark Hennessy]

I suspect one optocoupler is to take the PS25A power good signal back to the primary and the second one is from the error amp for the +5VSB. The TL594 should be on the primary side. I suspect this is not that an unusual design  although HP do odd stuff with connectors etc.

Yes, I did say that one is obviously for the standby supply - 12V in this case - but I'm not sure why a PG signal would go back from the secondary to the primary. That's all taken care of by the PFC section. That section runs all the time - even when the unit is in standby. If it didn't, then the SB rail would disappear. In standby, the PFC chip can run in pulse-skip mode to minimise the energy it uses. Of course, PG is generated for the motheboard, but the primary side doesn't need to know about that.

It's possible it's being used to shut down the supply in the event of an overload or over-voltage situation. I'm not seeing a conventional crowbar across the output - and those are a bit iffy given the high currents encountered in modern supplies - so these days OVP tends to implemented in a less brutal fashion. The TO220 in the middle of the PCB looked like it might have been an SCR (and D261 is conveniently placed to suggest it might be a zener associated with it), but I've got other ideas about what it might be...

But then, the shutdown could equally be achieved on the secondary side. So until I get my hands on one of these, that really is just a guess.

As for the TL594 being on the secondary side, that's actually very common and makes a lot of sense.

First, for a bridge converter, you need a drive transformer for the upper switch. So if going to the effort of specifying a transformer, you might as well drive both switches with it, even if the lower one could be done without. If nothing else, it's a zero-effort way to ensure both switches get the same waveform from the same source impedance - any waveform shaping required can be done once on the primary side of the transformer, rather than having to be done twice, perhaps in two different ways.

Having put the transformer there, you now have freedom of choice about where you put the controller IC.

If you put it on the primary side, you need voltage feedback from the secondary. That's just an optocoupler, but it's an optocoupler you wouldn't need otherwise.

Next, you have to power the controller IC. This means an additional winding on the main switching transformer. Commonly done, of course, but still extra cost.

That done, what about start up? You need to power the controller IC for the first few cycles before the supply establishes operation ("the engine fires"). Historically, this is usually done with a resistor dropping down the 300-something volts down to whatever the chip requires. This resistor wastes heat, so needs to be a large value (few hundred k), and the capacitor across the IC rails (the "car battery") needs to be in good condition so it can "turn the engine over" once it has been charged. Of course, high value resistors (car battery chargers) with lots of volts across them fail high, and capacitors (car batteries) fail. One or the other of these two components account for nearly all start-up faults in switched-mode power supplies...

With the controller on the secondary side, you clearly still have a start-up problem. One way to resolve that is to make the bridge on the primary side self-oscillating. Once the thing gets going, the controller IC takes charge. Neat. That's what the cheap Chinese PSU discussed in the other thread does.

Alternatively, you can put a completely separate supply in there. Luckily, we need a standby supply anyway, so it might as well power the TL594 while it's at it. So, that's a win-win. Lots of good reasons to put the TL594 on the secondary side, and lots of good reasons to not put in on the primary side.

What about the start-up circuit for the standby circuit?

While we don't know the part used in this example, most of these 7-pin ICs incorporate the start-up circuitry within the package. Usually it's a MOS-FET connected to the drain of the main switching MOS-FET. At power-up, the switching MOS-FET isn't being driven, so the unregulated DC input is present at its drain. The start-up MOS-FET passes a small current through to the control circuit and gets it going. When running, the start-up MOS-FET is turned off. No high-value resistors needed, and no need to send the raw DC into the IC on a separate pin. It's pretty smart IMHO...

Incidentally, those who remember the Thorn TX10 will be quite familiar with all this. This chassis was isolated - relatively unusual for the time - and had all the control on the secondary side. A pulse transformer carried the base drive, and a conventional 50Hz mains transformer powered the secondary side. Of course, the set had a standby mode, and most sets at the time that could be switched to standby used a 50Hz transformer for that purpose. Back then, we weren't so worried about standby power consumption.

All fascinating stuff. Well, I think so...

[Mark Hennessy]

I tested the PSU and there was some noise on receive, but this mostly went when I reconnected the PSU earth link, which I had lifted. Transmission reports come back with clear mod and no background noise, may need to confirm this with more local reports.
Also no indication of breakthrough on FM which I was getting before with the commercial Ham type PSU I bought and sent back.  The breakthrough on that was cured when I isolated the earth to the PSU, but I didn't wish to run the PSU with no earth and didn't want to invalidate warrantee by adapting the internal circuit to remove the mains earth from the DC output negative.

But I am happy to say the PSU seems suitable from RFI point of view, I think further improvement can be made on receive by earthing the PSU to the transceiver, better screening, and or adding more common mode filtering on the OP.
Its interesting to note that when the PSU is turned on there is a brief increase in noise on receive which goes after about 3 seconds.
Thanks
Mike

That's encouraging news. Shows it's worth persevering with this - I will do my best to borrow a PSU when I'm back at work tomorrow...

[Red to black]

One other point to bear in mind is that even fully floating SMPS (industrial ones are available) is that even though the mains "earth" is not connected directly to the secondary 0V line on these, there is usually a resistor between somewhere in the region of between either 2 X 4.7 Megohms in series or a single 10 Megohms , these are across the isolation barrier and are included to stop the secondary from floating up (or acquiring a charge) to a voltage higher than the withstand across the said barrier, also sometimes a high voltage low value Y cap is in this network (the cap is not present on all types).
It may be possible to disconnect the link between Mains Earth and the secondary side with the above in mind, however I am in no position to give any concrete re-design advice here.